The present disclosure provides a detection apparatus and method, and image processing apparatus and system. The detection apparatus extracts features from an image, detects objects in the image based on the extracted features; and detects key points of the detected objects based on the extracted features, the detected objects and a pre-obtained key point sets. According to the present disclosure, the whole detection speed can be ensured not to be influenced by the number of objects in the image to be detected while the objects and key points thereof are detected, so as to better meet the requirements of timeliness and practicability of the detection by the actual computer vision task.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A detection apparatus comprising: one or more processors; and at least one or more memories storing executable instructions which, when executed by the one or more processors, cause the detection apparatus to perform operations including: extracting features from an image; detecting an object in the image based on the extracted features; determining a pre-obtained anchor set corresponding to a category of an object based on the detected object; detecting a key point of the detected object based on the extracted features and the pre-obtained anchor set, wherein the key point represents a specific position of the detected object, and updating the pre-obtained anchor set by adding, to the pre-obtained anchor set, an anchor based on a specific shape of the detected object, wherein the pre-obtained anchor set is a set of a plurality of shapes included in the detected object of a specific category, and the key point indicates a position of the specific shape included in the plurality of the shapes.
2. The detection apparatus according to claim 1 , wherein in a case where the detected object is a human, the pre-obtained anchor set is an anchor set corresponding to a human body and the detected key point represents a part of the human body.
3. The detection apparatus according to claim 1 , wherein, the extracted features, the detected object and the detected key point are obtained by using a pre-generated first neural network.
4. The detection apparatus according to claim 1 , wherein executing the executable instructions causes the information processing apparatus to perform further operations including: selecting at least once on a part of key points among a plurality of detected key points of the object; and updating the selected key point of the object.
5. The detection apparatus according to claim 4 , wherein the selected key point is determined based on a predefined threshold and confidence information of the detected key points of the object.
6. The detection apparatus according to claim 1 , wherein in a case that the category of the object is a human, the pre-obtained anchor set includes a set of a plurality of shapes of human body, and the key point indicates a position of a specific human shape.
7. A detection method comprising: extracting features from an image; detecting an object in the image based on the extracted features; determining a pre-obtained anchor set corresponding to a category of an object based on the detected object; detecting a key point of the detected object based on the extracted features and the pre-obtained anchor set; and updating the pre-obtained anchor set by adding, to the pre-obtained anchor set, an anchor based on a specific shape of the detected object, wherein the pre-obtained anchor set is a set of a plurality of shapes included in the detected object of a specific category, and the key point indicates a position of the specific shape included in the plurality of the shapes.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
February 24, 2020
July 19, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.